Examination of CEFR-J spoken interaction tasks using many-facet Rasch measurement and generalizability theory

Attempts are underway to develop prototype tasks, based on a Japanese version of the Common European Framework of Reference for Languages (CEFR; Council of Europe, 2001; CEFR-J; Negishi, Takada, & Tono, 2013). As part of this larger project, the current paper reports on the creation of spoken interaction tasks for five levels (Pre-A1, A1.1, A1.2, A1.3, and A2.1). Tasks were undertaken by 66 Japanese university students. Two raters evaluated their interactions using a three-level holistic rating scale, and 20% of the performances were double rated. The spoken ratings were analysed using many-facet Rasch measurement (MFRM) and generalizability theory (G-theory). MFRM showed that all the tasks fit the Rasch model well, the scale functioned satisfactorily, and the difficulty of the tasks generally concurred with CEFR-J levels. Results from G-theory that employed the p x t design, including tasks as a facet, showed the different proportion of variance accounted for by tasks, as well as the number of tasks that could be required to ensure sufficiently high reliability. The MFRM and G-theory results effectively revealed areas for improving spoken interaction tasks; the results also showed the usefulness of combining the two methods for task development and revision.